Steer AI to compare history sources

Steer with format, constraints, and questions — not with “be smart.”

Format move: “Make a two-column table: Claim | Source A | Source B | How they differ.”

Example move: “Here is one difference I already noticed: Source A blames the tax; Source B focuses on representation. Add two more differences in the same style — do not write the essay yet.”

Question move: “Ask me three questions I should answer using my own notes before you draft anything.”

Each move keeps you in the essay. The AI becomes a comparison coach, not the author of record.

Paste only what you are allowed to use — short excerpts your teacher provided, or notes you typed yourself. Label them clearly:

Source A (class handout notes): … Source B (museum site summary I copied): …

Then ask:

List only disagreements about who, why, or what happened. Flag anything neither source supports.

That last line is gold. It stops the model from inventing a third “fact” that belongs to neither text.

Soft prompt: “Help me compare these sources.”

Steered prompt: “Using only the two labeled excerpts below, fill a table of disagreements. Quote a short phrase for each side. Do not add outside facts. End with two open questions I still need my notes to answer.”

The steered version is easier to check: every cell should point back to something you can underline in the sources.

Even a perfect table is raw material. Your essay still needs your thesis, your transitions, and your teacher’s required structure. If AI drafts full paragraphs, treat them as temporary scaffolding — rewrite in your voice or discard.

A good test: close the chat and explain one difference out loud in thirty seconds. If you cannot, you steered the model but not your learning.